大数据背景下基于改进RNN的低压配电网线损智能分析方法  

Improved RNN based intelligent analysis method for line loss of low-voltage distribution networks in context of big data

作　　者：李学军 张世元 LI Xuejun;ZHANG Shiyuan(College of Electrical Engineering and Automation,Fuzhou University,Fuzhou 350108,Fujian,China;Digitalization Division,State Grid Gansu Electric Power Co.,Ltd.,Lanzhou 730050,Gansu,China)

机构地区：[1]福州大学电气工程与自动化学院,福建福州350108 [2]国网甘肃省电力公司数字化事业部,甘肃兰州730050

出　　处：《沈阳工业大学学报》2025年第1期130-136,共7页Journal of Shenyang University of Technology

基　　金：国家自然科学基金青年基金项目(62001166)。

摘　　要：【目的】在电力系统中,线损率是衡量电网系统设计、运维和管理水平的重要经济技术指标,对于保障电网的稳定经济运行、提高供电效率具有重要意义。然而,在用户数量激增、用能特征多样化的大数据背景下,线损率的计算评价工作面临较大挑战。传统线损计算方法依赖于电网参数,精细化程度偏低,计算准确率不佳。【方法】针对该问题,提出了一种基于改进循环神经网络(RNN)的低压配电网线损智能分析方法,旨在通过智能化手段提高线损计算的准确性和效率。方法利用K-means算法对智能配电网的海量用户数据进行分类预处理,以降低数据冗余度。采用层次分析法(AHP)从分类数据中提取线损指标,这些指标随后被输入到深度学习模型中,其中,核心深度学习模型是由卷积神经网络(CNN)和改进长短时记忆网络(LSTM)模型融合而成,该模型能够挖掘配电网数据特征,实现线损的智能分析。通过IEEE33节点的仿真模型进行实验验证,充分展示所提方法的有效性。【结果】实验结果表明,所提方法的均方误差(MSE)和相对误差百分数(RE)分别为3.15 MW和2.43%,计算精度较高。与现有方法相比,所提方法在大数据背景下的配电网线损智能分析中具有明显优势,能够全面考虑各种配电网的影响因素,获得更精准的线损计算结果。此外,通过与两种经典文献方法进行对比实验,进一步验证所提方法的性能优势。【结论】基于改进RNN模型的低压配电网线损智能分析方法通过K-means算法和AHP预处理提取线损指标,再利用CNN-LSTM模型进行深入分析,有效提高了线损计算的准确性和效率。该方法主要针对低压配电网线路侧的线损进行分析,对于更高等级电压的线损分析尚未深入研究,但其在低压配电网线损智能分析中显示出优异的结果,具有实际应用价值。未来的研究将扩展到更广泛的校验分析,以[Objective]In power systems,the line loss rate is an important economic and technical indicator for measuring the design,operation,maintenance,and management levels of the power grid.It plays a significant role in ensuring the stable and economical operation of the power grid and improving the efficiency of power supply.However,in the context of big data characterized by a surge in the number of users and diversified energy usage characteristics,the management of line loss rates faces significant challenges.Traditional line loss calculation methods rely on power grid parameters,having a low level of refinement and poor calculation accuracy.[Methods]In response to these issues,this paper proposed an intelligent analysis method for line loss in low-voltage distribution networks based on an improved recurrent neural network(RNN),aiming to improve the accuracy and efficiency of line loss calculation through intelligent means.The method first utilized the K-means algorithm to classify and preprocess massive user data from smart distribution networks,thereby reducing data redundancy.Secondly,the analytic hierarchy process(AHP)was employed to extract line loss indicators from the categorized data,which were then input into a deep learning model.The core deep learning model was a fusion of convolutional neural networks(CNN)and an improved long short-term memory(LSTM)model,capable of mining distribution network data features to achieve intelligent analysis of line loss.The effectiveness of the proposed method was fully demonstrated through validation using an IEEE33-node simulation model.[Results]Experimental results indicate that the mean square error and relative error percentage of the proposed method are 3.15 MW and 2.43%,respectively,demonstrating high computational accuracy.Compared to existing methods,the proposed method has a distinct advantage in intelligent analysis of line loss in distribution networks in the context of big data,capable of comprehensively considering various factors influencing the distribution

关 键 词：大数据 低压配电网 K-MEANS聚类 层次分析法 特征提取 CNN-LSTM模型 智能线损分析 循环神经网络 

分 类 号：TM727[电气工程—电力系统及自动化]